A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

Apertus v1.1 0.5B Instruct — INT4

Ask questions, draft short text, or continue a conversation with a compact multilingual chat model. Returns an assistant reply; answers can be inaccurate and should be checked.

Browser inferencePowered by WebAI · Private · Unlimited runs
About this model

Ask questions, draft short text, or continue a conversation with a compact multilingual chat model. Returns an assistant reply; answers can be inaccurate and should be checked.

License
Apache-2.0
Model released
Unknown
Source repository created
May 1, 2026
Model weights · Q4
811.5 MB

Best for

  • Local multilingual chat, drafting, rewriting, and summarization.
  • Benchmarking browser LLM inference.

Keep in mind

  • Generated text may be incorrect, unsafe, biased, or fabricated and requires review.
  • The 0.5B parameter size limits reasoning depth, factual recall, and instruction following.
  • Accuracy depends on prompt language and input quality; uncommon languages and specialized domains may perform poorly.
  • The QAD INT4 conversion may reduce output quality, and browser backends may not support every candidate artifact.

Processed on your deviceInference runs in this browser. Input/output sharing requires consent for each run. Basic run logs include device information and approximate connection location. Model and runtime files download from external hosts. Saved downloads may be cleared by your browser.

Sources & technical details
Original model
swiss-ai/Apertus-v1.1-0.5B-Instruct
Release verification
Unknown
Release evidence
Exact original checkpoint public release date has not been verified. Repository creation, training, paper, artifact conversion and update dates are not release evidence.
Source updated
May 29, 2026
Browser model files
onnx-community/Apertus-v1.1-0.5B-Instruct-QAD-INT4-ONNX
Files updated
Jun 22, 2026
Artifact revision
defefbcf863cad09633955d5e82c8a1978f1c214

Input

CHAT & TEXT

Sample result

Apertus v1.1 0.5B Instruct — INT4

Settings

Limits the number of newly generated output tokens. Lower values finish sooner but can cut answers short. Higher values allow longer answers and can use more time and memory; they do not guarantee better answers.
Advanced settings 2
Adjusts randomness during token sampling. Lower values favor likely tokens and more repeatable wording. Higher values increase variation and can increase mistakes. It affects sampling only; some runtimes use zero for greedy decoding.
Chooses sampled rather than greedy token selection. On allows varied responses and makes sampling controls relevant. Off favors the most likely tokens for more repeatable responses; it does not guarantee correctness.
Estimated speedNot benchmarked

Output

Start a conversation

Send a message or try a sample. Replies appear here.

Benchmark evidence

Saved worker evidence is separate from benchmarks run on your device. It does not guarantee speed or quality on every browser.

Apertus v1.1 0.5B Instruct — INT4

4 configurations tested
Passed4
Failed0
TestedSep 9, 2026
RuntimeTransformers.js 4.2.0

Quality criteria

Exact-match answer, non-empty/non-repetitive output, and streamed text equal to final text.

PrecisionDeviceResult
Q4GPU · WebGPUAcceleratedPassedFast
Q4F16GPU · WebGPUAcceleratedPassedSlow
Q4F16CPU · WASMPassedVery slow
Q4CPU · WASMPassedVery slow
Test setup & sample details
Quality gate
Exact-match answer, non-empty/non-repetitive output, and streamed text equal to final text.
Tested
2026-09-09T16:50:14.459Z
Environment
Not recorded
Runtime
Transformers.js 4.2.0
Artifact revision
defefbcf863cad09633955d5e82c8a1978f1c214
Fixture
{
  "file": "apertus-v1.1-0.5b.json",
  "url": "https://f005.backblazeb2.com/file/hubters-webai-public/assets/benchmarks/apertus-v1.1-0.5b.json",
  "size": 397,
  "sha1": "01931912845892dd559d20d4f628d65a2e6999d8"
}
Rerun when
artifact revision change; runtime version change; fixture or quality gate change; browser/backend change
Per-case quality & failure details
[
  {
    "precision": "q4",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "exact-match",
      "output": "The capital of France is Paris.",
      "nonEmpty": true,
      "nonRepetitive": true
    },
    "runtime": {
      "loadMs": 67673.6,
      "inferenceMs": 5725,
      "totalMs": 75436.5,
      "firstChunkMs": 1929.3,
      "outputTokens": 7,
      "tokensPerSecond": 1.223,
      "grade": "very-slow"
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "exact-match",
      "output": "The capital of France is Paris.",
      "nonEmpty": true,
      "nonRepetitive": true
    },
    "runtime": {
      "loadMs": 71906.8,
      "inferenceMs": 333.3,
      "totalMs": 74056.1,
      "firstChunkMs": 265,
      "outputTokens": 7,
      "tokensPerSecond": 21.002,
      "grade": "excellent"
    }
  },
  {
    "precision": "q4f16",
    "device": "wasm",
    "status": "pass",
    "quality": {
      "grade": "exact-match",
      "output": "The capital of France is Paris.",
      "nonEmpty": true,
      "nonRepetitive": true
    },
    "runtime": {
      "loadMs": 46637.6,
      "inferenceMs": 5559.7,
      "totalMs": 53980.7,
      "firstChunkMs": 1970.8,
      "outputTokens": 7,
      "tokensPerSecond": 1.259,
      "grade": "very-slow"
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "grade": "exact-match",
      "output": "The capital of France is Paris.",
      "nonEmpty": true,
      "nonRepetitive": true
    },
    "runtime": {
      "loadMs": 50756.2,
      "inferenceMs": 854.7,
      "totalMs": 53346,
      "firstChunkMs": 501.1,
      "outputTokens": 7,
      "tokensPerSecond": 8.19,
      "grade": "slow"
    }
  }
]

Something not quite right?

Tell us what happened with Apertus 1.1. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.